biological control - darpa control proposers day (strychalski).pdfthe biological control program is...

Biological ControlDr. Elizabeth Strychalski, Ph.D.

PM, DARPA BTO

Proposers Day

22 February 2016

1Approved for Public Release, Distribution Unlimited

2Adapted from Menezes et al. Grand challenges in space synthetic biology. Interface (2015)

Nonbiological control Biology in the control loop

Biological control Hybrid control

Context: Control systems and biology

Approved for Public Release, Distribution Unlimited

3Adapted from Menezes et al. Grand challenges in space synthetic biology. Interface (2015)

Nonbiological control Biology in the control loop

Hybrid control

Context: Control systems and biology


Biological Control

The Biological Control Program is interested only in biological control

4

Control of biological systems using biological parts

Biological Control Program Vision: Enable the rational design of predictive, dynamic, and quantitative control of biological systems

© Shutterstock

hν

Environmental control

e-

Biological system


5



© Shutterstock

hν


e-

Biological system

Control strategy


6



Biological or nonbiological

inputs

Controlledsystem-level

biological outputs

© Shutterstock

hν

Closed-loop control systems made of biological parts to program behavior

across scales


e-

Biological controller

Biological system

Control strategy


7

Control with biology in the control loop today

Control engineering has been used to modulate human physiology through electronic medical devices

• Built using electrical and mechanical parts

• Designed using conventional control theory

• Embedded surgically

• Controlled at a single scale (e.g., molecular, cell, tissue, organ)

• Limited inputs and outputs

• Limited control of dynamics

Conventional control theory


8

Biological control tomorrow

Fight antibiotic resistance

New biotech platforms

Antifouling biofilm

BiologicalControl Program

Apply and advance control engineering to enable the rational design of biological control strategies for a variety of applications

• Build in control of biological systems using biological parts

• Design using a control theory suitable for biological systems

• Embed directly into and move with the biological system

• Operate across scales to control system-level function (e.g., genes to cells, cells to population)

• Multiple inputs and outputs

• Dynamic control strategies and outputs to account for environmental change and adaptation

• Leverage innate control in natural biological systems

Biologicalcontrol theory


© Shutterstock

9

New possibilities for engineering biology


© Shutterstock

ComplexityControl

© Shutterstock© Shutterstock

ComplexityControl

Biological Control

10

Terms for Biological Control BAA

Complexity• Refers to the high degree of interconnectedness of the scales and components of a biological system,

in a manner that complicates prediction and control at the state of the art

Generalizable control• Control strategies based on system-agnostic theory and modeling that, once well-understood and

implemented in one biological system, should be straightforward to adapt for use in other biological systems

System-level behavior• Complex behaviors targeted for control that manifest at the organizational level of a whole biological

system, for example, the unicellular organism, cellular community, multicellular organism, or ecosystem

• Includes, but is not limited to, growth and reproduction, adaptation and evolution, sensing and responding, and metabolism

Scales for Biological Control• Nanometers to centimeters, seconds to weeks, and biomolecules to populations of organisms

µs s hr d yr

nm µm mm m km

Non-living LivingApproved for Public Release, Distribution Unlimited

11

Notional schedule and structure

Phase 118 months

Phase 218 months

Phase 312 months

Technical Area 1: Controllers

Technical Area 2: Testbeds

Technical Area 3: Theory and Models

IV&V of Phase 1 and Phase 2 capabilities

Multidisciplinary Team

Must address all 3 Technical

Areas and PhasesNot part of

this BAA

Foundations For Control Of Biological Systems Applications And Generalizability


12

Technical Area 1 (TA1): Controllers

• Closed-loop control of biological systems built from biological parts only• Based on predictive theory and models, not ad hoc empirical screening• Inputs may be biological and/or nonbiological, while outputs must be biological• Control of system-level behaviors • Integrate multiple controllers into testbed individually and in combination

Phase 1 Phase 2 Phase 3Program

GoalsPredictable, closed-loop control of system-level behavior using a biological controller

Predictable, closed-loop control of additional system-level behaviors using additional biologicalcontrollers

Generalizable, predictable capabilities for biological control fora DoD application

TA1Goals

Assemble a biological controller described by TA3

Integrate into testbed in TA2

Build additional biological controllers described by TA3

Integrate into testbed in TA2

Work with TA3 to transfer to a DoD relevant demonstration system in TA2


13

Technical Area 2 (TA2): Testbeds• Consist of a biological system and hardware for quantitative measurements and dynamic

environmental control • Use a simple natural or synthetically-simplified biological system

• Examples include life-like systems, characterized microbes, and simple multicellular organisms

• Chosen for generalizability to additional biological systems and control strategies• Time-course measurements are required, and orthogonal measurements are encouraged





TA2Goals

Build and characterize testbed

Evaluate performance of controller from TA1

Share measurements with TA3

Integrate and evaluate additional controllers from TA1

Share measurements with TA3

Build DoD-relevant demonstration system

Integrate controllers from TA1 for practical biological control for a DoD application


14

Technical Area 3 (TA3): Theory and models

• Provide testable quantitative predictions for testbed and controllers• Provide models and prediction for biological controllers and control strategies, as opposed to

merely describing experimental results• Adaptable and generalizable for modeling control of biological systems beyond the testbed• Should include the development of new control theory and modeling approaches, as needed





TA3Goals

Develop predictive models for control for TA1 and TA2

Exchange domain knowledge with TA1and TA2

Inform development of additional controllers and control strategies in TA1

Implement models in a computational environment for rational design of control

Translate control strategies for a DoDapplication in TA1 and TA2

Participate in tests of the models’ predictive capabilities


15

Independent Validation and Verification (IV&V)

Controllers should behave reproducibly, regardless of laboratory, location, operator, etc. • The Biological Control Program will employ IV&V

Proposals must include plans to deliver all necessary materials, protocols, and domain knowledge to the IV&V team(s)

IV&V team(s) will be selected separately from BAA-16-17

If duplication of the testbed presents an unreasonable cost, teams should outline a plan for IV&V team(s) to access the team’s laboratory for validation, verification, and testing

Predictive capabilities developed in TA3 will be tested in Phase 3, in a manner to be announced by DARPA no later than the end of Phase 2


16

Phase 3: DoD relevant application

Demonstrate the utility of biological control in applications of relevance to DoD

• Proposer-defined applications with impact for national security

• Scientific proof-of-concept for control of biological systems at the laboratory scale

• Based on capabilities from Phase 1 and Phase 2

Candidate Applications


17

Metrics and milestones: Specified and proposer-defined

Few specific metrics and milestones are defined in BAA-16-17, to allow flexibility and creativity

Teams must fill this gap with clear proposer-defined metrics and milestones, as appropriate• Justify these relative to the state of the art• Explain how these build towards the overall goal of the proposal and program

Clear, concise metrics and milestones demonstrate the strength of your proposal• Strong preference for quantitative metrics over qualitative metrics• Meant to quantify and evaluate progress and accomplishments of proposed

work• Central to contracting


18

BAA timeline and funding overview

Abstracts• Strongly encouraged• Abstract due date: March 18, 2016• Purpose is to determine if the idea fits within the program before

undertaking the effort of a full proposal

Full proposal due date: April 29, 2016

Available Biological Control funds• Up to approximately $38M will be awarded • Number of teams selected depends on the quality of proposals and

availability of funds


19

Teaming

Teaming is strongly encouraged• Require deep expertise in both theoretical and experimental science and

technology, for example, in fields of biology relevant to the proposed testbedand demonstration system, fields of engineering relevant to the proposed testbed and measurement methods, and control engineering

Equal emphasis on experiment and theory

Address all Phases and TAs

Designated PI or Team Coordinator will be the primary interface with DARPA

The Biological Control program has no bias for teams internal to one institution or across multiple institutions• Plan for effective communication and collaboration within teams


20

Communicating research outside the program

Biological Control is funded as basic research

Performers are not expected to be censored• DARPA does not anticipate a requirement for review of research prior to

publication• Individual contracts will detail review requirements

DARPA will request two-week notice and copies of presentation slides, manuscripts, and other communication prior to distribution


21

BAA inbox and FAQ

Direct all questions and communication to the BAA inbox• [email protected]• Elizabeth, any member of her team, or any member of the scientific review

panel will not communicate directly with a potential proposer regarding BAA-16-17

• All communication will occur through the BAA inbox only• Elizabeth and the BAA inbox will provide feedback and/or guidance to clarify

content of BAA-16-17 only and cannot provide feedback regarding any aspect of a proposal

BAA inbox FAQ• DARPA will post a consolidated FAQ regularly• Access the posting at:

• http://www.darpa.mil/work-with-us/opportunities/Solicitations/BTO_Solicitations

• Submit questions to [email protected] at least 15 days prior to the proposal submission deadline


http://www.darpa.mil/work-with-us/opportunities/Solicitations/BTO_Solicitations

22

Advice

Read the BAA carefully and thoroughly

Ask for clarification• FAQs will be updated regularly

Refer to material from Proposers Day, which will be available online

Be bold, take risks, add urgency to forbearance, and have fun

Questions?


www.darpa.mil

23Approved for Public Release, Distribution Unlimited

24

Additional Slides


25

Genotypes

Phenotypes

Syst

em-le

vel f

unct

ion

A mechanical or electrical system is designed using conventional control theory to program and maintain a desired, predictable state• Rational “blueprint” from theory to

system implementation for a conventional control system made from electrical and mechanical parts

A natural biological system maintains its state, but the control theory and mechanisms to program control across scales are largely unknown • Incomplete “blueprint” connecting

genotype to system-level function is largely unknown for a biological system made from biological parts

Adapted from Biological Evolution and Statistical Physics, Peter Stadler (2002)

Unknown mappingacross scales

? ?

Biological control ≠ electrical, mechanical control


Existing control theory must be advanced to address the unique set of challenges presented by biological systems

26

Biological systems as control systems

All control systems, including natural biological systems

Conventional control theory provides a foundation for the rational design of predictable biological control

• Measure internal state • Compare against a desired state• Adjust state accordingly

Conventionally-controlled system• Models for prediction• Clear causality• Full knowledge of parts• Full knowledge of organization

Biological system• Stochasticity and complexity• Unknown causal relationships• Unknown/Undefined parts• Unknown/Undefined interactions

Centrifugal governor: Mechanical control system that uses negative feedback to regulate rotational speed

Adapted from Hasty et al. (2001)Repressor/protein fusionPromoter

Repressor: Autoregulatinggene circuit that uses negative feedback to regulate rate of protein production

Example of control:Negative feedback


Technical Areas

State of the art (SOA) Biological Control

Deliver predictive, embedded biological controllers with tunable, multiscale output for rational design

Deliver models with a system

approach

Simple models with a parts approach

TA3Biological control

engineeringL{y(t)} = f(t)

27

Build simple testbeds

Complex biological testbeds

TA2Measurements

for predictability across scales

Few parts act at single scale TA1

Integrate many biological parts across scales

Multiscalecontrollers

New approaches


TA2Testbed for biological control

TA3Theory and models

TA1Biological controller

Integrated Technical Areas (TAs)

Biological control engineering

28

Measurements for

predictability across scales

Assemble biological controllers


29

Community for biological control


• Synthetic biology• Ecology• NEMS/MEMS• Biosensors

• Control theory• Complexity science• Systems engineering• Biophysics

•Recognized need•DARPA investment•Teaming

ScienceTechnology

Theory Modeling

Community

E. coli

Integrated biological parts and multiscale signaling

Adapted from Moon et al. (2012)

Integration exists to combine engineered signals but not coupled across scales to control system-level function

30

Inputs Sensors Circuits Output

Slime moldDictyostelium

Unicellular Multicellular

Small moleculesignaling

Thomas Gregor

Must sense all 4 inputs to generate response

Red fluorescent

protein

ANDgate

ANDgate

ANDgate

Senseinput 3

Senseinput 4

1

2

3

4

Senseinput 1

Senseinput 2


TA1 solution: Biological controllers

nm µm

Rein & Deussing (2012)

Orelle et al. (2015)

Combine biological parts rationally for controlled function across scales

Integrate

31

Example: Assemble controller for biofilm growth through selective cell death

Genetic circuitfor programmability

Fluorescentprotein readout

Optogeneticsfor tunability

Synthetic ribosome for modularity

Synthetic virus forintercellular signaling

Citorik et al. (2014)


TA2 SOA: Genetically minimized cell testbed

© JCVI unpublished (2015)

Minimized cell is simplest living testbed but does not allow study oflarger-scale, multicellular functions

32

477 genes

~985 genes

M. mycoides

Genome reduced by half

Before

After

Biosynthesis of cofactors & prosthetic groups Cell envelopeCellular processesCentral intermediary metabolismDNA metabolismEnergy metabolismFatty acid and phospholipid metabolismHypothetical proteinsNoncoding RNA featureProtein fateProtein synthesisPurines, pyrimidines, nuc’sides & nuc’tidesRegulatory functionsSignal transductionTranscriptionTransport and binding proteins

Modularize cell functions

Energy metabolism

Protein fateProtein synthesis

Signal transduction

Transport and binding proteins


nm µm mm cm

TA2 solution: Testbeds to measure biological control

Develop tractable lab-scale testbeds with controlled environments for high-quality measurements of controller function

PredictIntegrate

33

Building tractable testbeds• Experimental

protocols• Theoretical

understanding• Simplify to

essential functions

Example testbedsHydraGenetically

minimized cell

© JCVI (2015)

Cell-free system

Gibson (2014) © Yuste (2015)

• Molecular scale• Gain mechanistic

understanding

• Cellular scale• Connect genotype

to phenotype

• Organism scale• Connect molecular

to multicellular scales


Buffered: Predictable, load-independent output

Performancerestored

TA3 SOA: Theory for rational design of control

34

Yeast

Adapted from Mishra et al. (2014)

Adapting strategies from control theory aids predictability for rational design of control but has not been scaled-up

Slow dynamics Fast dynamics Slow dynamics

Phosphorylation cascadebuffer

Signaling protein production

Reversible protein binding

load

Fluorescent protein

readout

Signaling protein production

Reversible protein binding

load

Unbuffered: Load-dependent output

≤ 75 % lag in response timeFluorescent

proteinreadout


TA3 solution: Adapt existing control theory for controlled biological solutions to technological problems

Develop theory and models to design a variety of biological controllers and evaluate controller function

Adapted from Harris et al. (2015)

Biological description

Σ SystemInput Output

Control analysis• Performance• Stability• Robustness

Predict Control

35

Example: Biological integrator for ideal adaptation to environmental

change

Translate variety of control strategies, e.g.• Open-loop• Closed-loop• Feedback• Feedforward

Abstract control description


FOR OFFICIAL USE ONLY (FOUO) – NOT FOR PUBLIC RELEASE© Shutterstock

Grow

Sense

Adapt

Metabolize

Inputs

Outputs

Input

Output

Biological System

Sensor

Controller

Embedded control using biological parts, e.g.• Synthetic DNA/RNA circuits• Proteins• Synthetic organelles/cells

Predictive theory to build in tunable control

hν

Measurements and environmental control• Predictability• Reproducibility• Generalizability

To help protect your privacy, PowerPoint has blocked automatic download of this picture.

Biological control

e-

Enable the rational design of predictable, dynamic, and quantitative control for biological systems

Closed-loop control across scales• Space• Time• Biological organization

Biological control using biological parts


Biological system

Biological or nonbiological inputs

biologicaltoolbox

37

Control Biological Systems Using Biological Parts

Foundations for multiscale control of biological systems

© Shutterstock

Closed-loop control systems made of biological parts to program behavior

across scales

Biological controller


Grounded in theory and mathematical modeling

Characterized rigorously in simplified testbeds

Inspired by conventional control engineering

Constructed from biological parts

biological control - darpa control proposers day (strychalski).pdfthe biological control program is...

Documents